MDM2 Mediates p300/CREB-binding Protein-associated Factor Ubiquitination and Degradation*

We recently reported that MDM2, a negative feedback regulator of the tumor suppressor p53, inhibits p300/ CREB-binding protein-associated factor (PCAF)-medi-ated p53 acetylation. Our further study showed that MDM2 also regulates the stability of PCAF. MDM2 ubiquitinated PCAF in vitro and in cells. PCAF ubiquitination occurred at the N terminus and in the nucleus, as the nuclear localization signal sequence-deletion mutant of MDM2, which localized in the cytoplasm and degraded p53, was unable to degrade nuclear PCAF. Restriction of PCAF in the nucleus by leptomycin B did not affect MDM2-mediated PCAF degradation. Consis-tently, overexpression of MDM2 in p53 null cells caused the reduction of the protein level of PCAF, but not the mRNA level. Conversely, PCAF levels were higher in MDM2-deficient mouse p53 (cid:1) / (cid:1) / mdm2 (cid:1) / (cid:1) embryonic fibroblast (MEF) cells than that in MDM2-containing MEF cells. Furthermore, MDM2 reduced the half-life of PCAF by 50%. These results demonstrate that MDM2 regulates the stability of PCAF by ubiquitinating and degrading this protein. Ubiquitin-dependent proteolysis plays a pivotal role in reg-ulating the turnover of many proteins that are crucial for cell homeostasis and growth. One important ubiquitin-proteasome pathway closely related to cell growth

Ubiquitin-dependent proteolysis plays a pivotal role in regulating the turnover of many proteins that are crucial for cell homeostasis and growth. One important ubiquitin-proteasome pathway closely related to cell growth and tumorigenesis is mediated by an E3 ubiquitin ligase protein called MDM2 (1)(2)(3)(4). MDM2 was originally identified as an oncoprotein (5,6), and its levels are increased in 33% of human sarcomas and 50% of human leukemias (7,8). The oncogenic activity of MDM2 is attributed primarily to its inhibitory effect on the function of the tumor suppressor p53 protein (9,10). p53 is a nuclear transcriptional activator that induces the expression of many genes, protein products of which mediate p53-dependent cell growth arrest and apoptosis, thus suppressing cell transformation and tumorigenesis (11,12). Normally, p53 is hardly detectable in most tissues or cells because of its rapid degradation mediated by MDM2 (1)(2)(3)(4). Stress signals turn on various cellular pathways that protect p53 from targeting by MDM2 and thus stabilize and activate p53 (13)(14)(15)(16)(17). Because active p53 is extremely toxic to cells, MDM2 is also overproduced through transcription activation by p53 to monitor p53 activity in response to stresses (10,18,19). Thus, MDM2 serves as a nega-tive feedback regulator of p53 as well (19 -21). The physiological significance of this feedback regulation is substantially illustrated by the fact that the lethal phenotype of the mdm2 knock-out mice is rescued by further deleting the p53 gene (22,23). Hence, p53 is the prime target of the MDM2-mediated ubiquitin-proteasome system in cell growth regulation.
Whether MDM2 targets other proteins for ubiquitination and degradation has been a tempting question. An increasing number of proteins involving cell growth control, such as p14 arf (p19 arf for the mouse homolog), p300, CBP, Rb, E2F1, ATM, cyc G, PP2A, AKT, p73, and p63, have been shown to physically interact with MDM2 (24 -39). These proteins are either upstream regulators of MDM2 or are regulated by MDM2. However, no evidence has been presented so far to suggest that MDM2 could ubiquitinate these proteins or reduce their stability. However, several other proteins have been identified recently as the substrates for the MDM2 E3 ubiquitin ligase. These proteins include MDMx, ␤-arrestin, G-protein-coupled ␤ 2 -adrenergic receptor, androgen receptor, glucocorticoid receptor, and Tip60, a histone acetyltransferase (40 -44). Recently, we found that MDM2 can lead to p21 degradation in a Ring finger domain-independent manner (58) and also ubiquitinate PSD95 (59). Identification of these non-p53 protein targets for MDM2 indicates that MDM2, in addition to regulating the p53 pathway, also plays a role in different cellular processes. Thus, identifying unknown substrates of MDM2 would reveal its new functions in cells.
In our attempt to identify other possible substrates of the E3 ubiquitin ligase MDM2, we have tested whether a p300/CREBbinding protein-associated factor (PCAF) 1 is a candidate. The idea for this test stemmed from our recent report in which we showed that MDM2 interacted with PCAF and inhibited its acetyltransferase activity on p53 (45). PCAF contains an intrinsic histone acetyltransferase and functions as a coactivator for a number of transcriptional activators (46,47). PCAF also acetylates p53 at lysine 320 in response to DNA damage signals and activates the activity of p53 (48). Therefore, it is significant if MDM2 can target PCAF for degradation. Indeed, our study as described here demonstrates that MDM2 ubiquitinates PCAF in vitro and in cells, leading to its degradation.
Purification of the PCAF-associated Complex-The PCAF-associated complex was purified from stable Flag-PCAF-expressing HeLa cells using the method established previously (53). Purified proteins were analyzed by silver staining and immunoblotting assays.
Transient Transfection and Immunoblot Analysis-H1299, HEK 293, and 293-HA-MDM2 stable cell lines (60% confluence in a 60-mm plate) were transfected with PCAF (1 g) either alone or together with pCDNA3-HA-MDM2 or deletion mutant MDM2 (see the figure legends for the amount of plasmids used). 48 h post-transfection, cells were harvested for the preparation of whole cell lysates. Whole cell lysates containing 50 or 100 g of protein were directly loaded onto an SDS gel, and proteins were detected by ECL reagents (Bio-Rad) after immunoblotting using antibodies as indicated in the figure legends.
Immunoprecipitation-Immunoblot Analysis-Transfected HEK cells were harvested for preparation of cell lysate containing 300 g of proteins used for immunoprecipitation, followed by immunoblot as described previously.
Northern Blot Analysis-Northern blot analysis was conducted as described (54). Total RNA was isolated from transfected H1299 cells using the TRIzol reagent (Invitrogen). Fifteen g of RNA were loaded onto a 1.5% agarose gel and transferred to a nitrocellulose membrane. The membrane was exposed to UV light in a UV cross-linker (Fisher Biotech) and incubated with 32 P-labeled cDNA probes encoding human PCAF at 42°C overnight. After washing with 4ϫ SSC once and 1ϫ SSC twice, the blot was exposed to x-ray film overnight. Analysis of PCAF Half-life in Cells-For determination of the exogenous PCAF half-life, mouse p53 Ϫ/Ϫ /mdm2 Ϫ/Ϫ embryonic fibroblast cells or H1299 cells were transfected with the Flag-PCAF plasmid alone or together with the MDM2 expression plasmid as described above. 48 h after transfection, transiently transfected cells in 100-mm plates were treated with 50 g/ml cycloheximide and harvested at different time points as indicated for the preparation of cell lysates. Equal amounts of proteins were analyzed by SDS-PAGE followed by immunoblot using the monoclonal anti-Flag antibody. These experiments were repeated twice.
PCAF in Vitro Ubiquitination Analysis-The MDM2-mediated in vitro ubiquitination assay was performed as described previously (3) with some modifications. The PCAF in vitro ubiquitination assay was carried out in 20 l of a reaction mixture containing 50 mM Tris-HCl (pH 8.0), 5 mM MgCl 2 , 0.5 mM dithiothreitol, 2 mM NaF, 3 M okadaic acid, 100 ng of Flag-PCAF, 100 ng of MDM2, 200 ng of UBA-1, 200 ng of UbcH5b, 500 ng of HA-ubiquitin, 5 mM ATP (Amersham Biosciences), 1.5 mM ATP␥S (Fisher ICN). As a control, His-p53 was used instead of PCAF. The mixture was incubated at 37°C for 60 min and analyzed on SDS-PAGE afterward. Ubiquitinated PCAF was detected by immunoblot using the polyclonal anti-PCAF antibody. Ubiquitinated His-p53 was detected by immunoblot using the monoclonal HA antibody.

FIG. 1. Establishment of in vitro
MDM2-dependent ubiquitination assays. A, purification of E1 (UbA-1) and E2 (UbC-H5B). E1 and E2 were purified as described under "Experimental Procedures." Purified proteins (1 g of E1 and 5 g of E2) were analyzed by SDS-PAGE followed by Coomassie Blue staining. Molecular markers are indicated on the left, and proteins are marked on the right (the same is true for the following figures). B, purification of HA-ubiquitin. HA-ubiquitin was purified as described under "Experimental Procedures" and analyzed by SDS-PAGE followed by Coomassie Blue staining (C.S.) and immunoblot analysis. C, purification of MDM2 and p53. MDM2 and p53 were purified as described in the text and analyzed by SDS-PAGE followed by silver staining (for MDM2) or Coomassie Blue staining (for p53) analysis. D, in vitro MDM2-mediated p53 ubiquitination. This assay was performed as described in the text. The amount of p53 used was 100 ng. 1x and 2x of MDM2 denote 100 ng and 200 ng. 1x and 2x of HA-Ub indicate 250 and 500 ng of HA-ubiquitin. NaH 2 PO 4 , 10 mM Tris-HCl, pH 6.3, 10 mM ␤-mercaptoethanol). Proteins were eluted from beads with buffer D (200 mM imidazole, 0.15 M Tris-HCl, pH 6.7, 30% glycerol, 0.72 M ␤-mercaptoethanol, and 5% SDS). The eluted proteins were analyzed by immunoblot for the polyubiquitination of PCAF with monoclonal Flag antibodies (Sigma).
Immunofluorescent Staining and Fluorescent Microscopic Analysis-H1299 cells were co-transfected with the plasmid encoding PCAF alone or with the MDM2 or ⌬50 -230MDM2 expression plasmids. 48 h after transfection, cells were fixed for immunofluorescent staining with monoclonal anti-Flag antibodies and polyclonal anti-MDM2 antibodies as well as for DNA staining with DAPI. The Alexa Fluor 488 (green) goat anti-mouse antibody and the Alexa Fluor 546 (red) goat anti-rabbit antibody (Molecular Probes) were used for p21 waf1/cip1 and MDM2, respectively. Stained cells were analyzed under the Zeiss Axiovert 25 fluorescent microscope.
Glycerol Gradient Sedimentation Centrifugation-H1299 cells (60% confluence/10-cm plate ϫ 3 plates) were transfected with the PCAF plasmid either alone or together with the MDM2 plasmid. Cells were

FIG. 2. MDM2 ubiquitinates PCAF in vitro and in cells.
A, MDM2 is the E3 ligase for PCAF ubiquitination in vitro. Reactions contained components as shown in the figure and were performed as described under "Experimental Procedures." Ubiquitinated PCAF was detected with anti-PCAF antibodies. The symbol * indicates unknown bands. B, purification of Flag-PCAF. Flag-PCAF was purified from the baculovirus system using agarose beads coupled with anti-Flag antibodies. Purified proteins were analyzed using silver staining (left) and immunoblot with the polyclonal anti-PCAF antibody (right). C, MDM2 mediates PCAF polyubiquitination in cells. HEK 293 or 293 HA-MDM2 cells (60% confluence/ 10-cm plate) were transfected with plasmids encoding His-ubiquitin (2 g), pCX-Flag, or pCX-Flag-PCAF (2 g for X). Transfected cells were treated with 10 M proteasome inhibitor MG132 overnight before being harvested. The in vivo ubiquitination assay was carried out as described under "Experimental Procedures." Polyubiquitinated Flag-PCAF was pulled down by Ni-NTA beads and detected by IB with anti-Flag antibodies (top panel). The level of Flag-PCAF was detected by direct IB analysis (second and third panels from the top; L.E. and S.E. denote longer and shorter exposures, respectively). Polyubiquitination of MDM2 was probed on the same blot as that described above using anti-MDM2 antibodies (second panel from bottom). MDM2 levels were detected by direct IB analysis. D, the 60 -464 amino acid region of PCAF is the target for ubiquitination by MDM2. The schematic presentation of the PCAF deletion mutants is shown at the top of the data (⌬yGCN5 denotes the region homologous to yeast GCN5). HEK 293 cell were transfected with 4 g of plasmids encoding pCX-Flag-PCAF, pCX-Flag-⌬yGCN5 PCAF (lanes 4 and 5), or pCX-Flag-⌬0 -464 PCAF (lanes 6 and 7) alone or together with 2 g of the pET-His-ubiquitin plasmid or pCX-Flag plasmid. Transfected cells were treated with 10 M MG132 overnight before being harvested. Polyubiquitinated PCAF molecules were purified on Ni-NTA beads and detected by IB with antibodies against Flag (upper panel). Direct IB was conducted using anti-Flag antibodies (lower panel). * indicates the nonspecific signals caused by cross-reactions with this antibody. harvested 48 h after transfection for preparation of whole cell lysates. 400 g of lysates from each sample were loaded onto the surface of a 12.5-25% glycerol gradient buffer in a 12-ml centrifuge tube. The samples were subjected to centrifugation in a SW41TI rotor (Beckman) at 32,000 rpm for 20 h. Approximately 250 l per fraction were collected from each tube after centrifugation, and 30 l of each fraction were used for IB analysis as shown (see Fig. 6B).

Establishment of an in Vitro MDM2-mediated Ubiquitination Assay with Highly
Purified Proteins-To identify new substrates for MDM2, first we wanted to establish an in vitro MDM2-mediated ubiquitination reaction using a defined system with purified proteins. As shown in Fig. 1, A-C, all of the required recombinant proteins including E1 (UbA-1), E2 (UbC-H5B), HA-ubiquitin, and MDM2 were highly purified. As a positive control, highly purified p53 (Fig. 1C) was used to test the activity of these pure proteins. In the reconstituted ubiquitination reaction, p53 was ubiquitinated by MDM2 in a dosedependent manner (Fig. 1D). This ubiquitination was dependent on MDM2 (lane 1) and HA-ubiquitin (lane 6), as p53 was not ubiquitinated in the absence of either of them. It was also dependent on E1 and E2 (data not shown) as expected (3). This result was reproducible and clearly demonstrated that we were able to establish an efficient MDM2-dependent ubiquitination assay in vitro using highly purified proteins.
MDM2 Ubiquitinates PCAF in Vitro and in Cells-After establishing the in vitro MDM2-mediated ubiquitination assay, we tested whether MDM2 was able to ubiquitinate PCAF in vitro. This idea originated from our recent study showing that MDM2 bound to PCAF and inhibited PCAF-mediated p53 acetylation and activation (45). To test this idea, Flag-PCAF was purified from the baculovirus expression system as shown in Fig. 2B. Using this purified PCAF, an in vitro PCAF ubiquitination assay was performed with the purified proteins as described above (Fig. 1). As shown in Fig. 2A, PCAF was ubiquitinated in vitro. This ubiquitination was dependent on MDM2, as no apparent ubiquitination of PCAF was detected in the absence of MDM2 (compare lane 4 with the rest of the lanes). This result was reproducible and suggests that MDM2 can ubiquitinate PCAF in vitro.
To further confirm that PCAF is ubiquitinated by MDM2 in cells, we conducted an in vivo PCAF ubiquitination assay using HEK 293 cells, which contain a low level of MDM2, and stable MDM2-overexpressed HEK 293 cells (bottom panel of Fig. 2C). These cells were transfected with plasmids encoding His-ubiq- uitin or Flag-PCAF alone or together. Transfected cells were treated with the proteasome inhibitor MG132 prior to being harvested for preparation of cell lysates. Ubiquitinated proteins were isolated using Ni-NTA beads and detected with antibodies against Flag-PCAF and MDM2. As shown in To further determine which region of PCAF is required for ubiquitination by MDM2, the same in vivo ubiquitination assay using the stable MDM2expressed 293 cell line was performed, except that two PCAF deletion mutants without either the N terminus or C terminus were used as controls. As shown in Fig. 2D, the N-terminal half of PCAF lacking the GCN5-like domain was ubiquitinated in the MDM2-expressed 293 cells, whereas the C-terminal fragment lacking the region of amino acids 60 -464 was not ubiquitinated even in the presence of His-ubiquitin (compare lane 5 with lane 7). The result indicates that the ubiquitination occurs in the N-terminal domain of amino acids 1-493, which is consistent with our previous study showing that this domain bound to MDM2 directly (45). Taken together, these results demonstrated that MDM2 ubiquitinates PCAF in cells, and the ubiquitination occurs at the MDM2-binding N terminus of PCAF.
MDM2 Ubiquitinates the Complex-associated PCAF in Cells-Because PCAF has been shown to associate with other proteins including hTAF II 31, forming a large multisubunit complex in cells (53,56), we wanted to test whether MDM2 is able to target the complex-associated PCAF. First, we determined whether MDM2 could interact with the complex-associated PCAF by comparing the PCAF-associated complex with the Flag-PCAF monomer. The former was purified from a stable Flag-PCAF-expressing HeLa cell line (Fig. 3, A and B), and the latter was purified from insect cells (Fig. 2B) through a Flag-immunoaffinity column followed by gel-filtration chromatography (data not shown). As shown in Fig. 3C, PCAF was  5. MDM2 leads to a decrease of the PCAF level in the nucleus. A, MDM2 leads to a decline in the protein level but not the mRNA level of PCAF in cells. H1299 cells (60% confluence/10-cm plate) were transfected with pCX-Flag-PCAF plasmid (2 g) and pCMV-HA-MDM2 plasmid (x denotes 2 g) alone or together. Cells were harvested 48 h after transfection. 100 g of proteins were loaded onto an SDS gel followed by IB using antibodies as indicated on the left. 15 g of total RNA were used for Northern blot (two bottom panels). B, MDM2, but not its NLS deletion mutant, leads to a decrease of PCAF in cells. HEK293 cells were transfected with pCX-Flag-PCAF plasmid (2 g), pCMV-HA-MDM2 plasmid (4 g), and pHDM-⌬NLS (⌬50 -230) plasmid (4 g) either alone or together with the Flag-PCAF plasmid. Cells were harvested 48 h after transfection for the preparation of cell lysates. 300 g of proteins from each sample were analyzed by immunoprecipitation with monoclonal anti-Flag antibodies followed by IB analysis with antibodies against MDM2 (top panel) or PCAF (middle panel). The bottom panel shows the expression levels of wild type and deletion mutant MDM2 proteins as detected by straight IB with anti-MDM2 antibodies. C, immunofluorescent staining of MDM2 in cells. 60% confluent H1299 cells/60-mm plate were transfected with pCX-Flag-PCAF plasmid (1.5 g) together with pCMV-HA-MDM2 plasmid (1.5 g) or pHDM-⌬NLS (⌬50 -230) MDM2 plasmid (1.5 g). Immunofluorescent staining of Flag-PCAF and MDM2 was conducted 48 h after transfection as described. Polyclonal anti-MDM2 and monoclonal anti-Flag antibodies were used to detect MDM2 or Flag-PCAF as indicated on top of each image. DAPI staining reveals the nucleus. D, MDM2 mediates PCAF degradation in the presence of the nuclear export inhibitor leptomycin B. H1299 cells were transfected with plasmids as indicated above the panel. Cells were treated with or without 15 ng/ml leptomycin B (LMB) 24 h post-transfection and harvested overnight after the treatment. 150 ng of total proteins were loaded to an SDS gel for IB analysis using antibodies as indicated on the left. Methanol (MeOH) was used to dissolve LMB and also was used here as a control. E, the proteasome inhibitor MG132 prevents PCAF reduction by MDM2. H1299 cells were transfected with pCX-Flag-PCAF and/or pCMV-HA-MDM2 plasmids and treated with Me 2 SO (lane 2) or 10 M MG132 (lane 3) overnight 36 h after transfection. 100 g of proteins from each cell lysate were used for IB analysis with antibodies as indicated. F, PCAF levels are inversely proportional to MDM2 levels in MEF cells. 100 g of proteins from p53 Ϫ/Ϫ or p53 Ϫ/Ϫ mdm2 Ϫ/Ϫ MEF cell lysates were analyzed by IB with antibodies as indicated.
co-immunoprecipitated with MDM2 by the anti-MDM2 antibody when either the Flag-PCAF-associated complex or the PCAF monomer was used. The observation that MDM2 can interact with PCAF even when it is in the complex suggested that MDM2 may target the complex-associated PCAF for ubiquitination as well. To test this idea, we performed an in vivo ubiquitination assay. Stable Flag-PCAF-expressing HeLa cells, which were used for purifying the PCAF-associated complex (Fig. 3), were transfected with MDM2 and His-ubiquitin plasmids and harvested 48 h after transfection. The Flag-PCAFassociated complex was purified through the same two-step procedures as those used for the results of Fig. 3: Flag-immunoaffinity and gel filtration columns. Consistent with the results of Fig. 3, the purified PCAF associated with a complex of ϳ700 kDa (Fig. 4A). Interestingly, this PCAF protein was also ubiquitinated (Fig. 4B). This ubiquitination was indeed mediated by MDM2, as no ubiquitination of PCAF was detected in the absence of MDM2 even though His-ubiquitin was present (Fig. 4, C and D). Because Flag-PCAF already associated with the complex (Fig. 3A) prior to the transient introduction of MDM2 into the cells, ubiquitination of PCAF by MDM2 must occur when PCAF is in the complex instead of in a free form. These results demonstrate that MDM2 can bind to and ubiquitinate PCAF even when PCAF is in the complex.
MDM2 Leads to the Decrease of PCAF Level in Cells-Next, we wanted to determine whether MDM2 regulates the stability of PCAF. To test this idea, human non-small cell carcinoma H1299 cells, which are devoid of p53, were transfected with the Flag-PCAF plasmid and HA-MDM2 plasmid either alone or together. Cells were harvested 40 h after transfection for immunoblot and Northern blot analyses. As shown in Fig. 5A, overexpression of MDM2 led to a marked decrease in the PCAF level (top panel). This decrease was not caused by the inhibition of PCAF transcription, as its mRNA level was not affected in the presence of MDM2 (third panel from the top). To test whether MDM2 affects the level of PCAF in the nucleus or cytoplasm, a similar transfection assay was conducted except that an MDM2 deletion mutant lacking the NLS-containing region was used. Proteins were analyzed by immunoprecipitation with anti-Flag antibodies, followed by immunoblot analysis with antibodies against PCAF and MDM2, respectively. As shown in Fig. 5B, overexpression of wild type, but not the deletion mutant, MDM2 resulted in the drastic decline of the PCAF level. Also, wild type, but not the mutant, MDM2 was co-immunoprecipitated with anti-Flag antibodies, suggesting that the mutant MDM2 did not interact with Flag-PCAF. This result is in agreement with our previously published results (45) and also is consistent with the fact that the NLS-deleted MDM2 mutant localized in the cytoplasm whereas PCAF stayed in the nucleus (Fig. 5C). Because this deletion mutant was still able to mediate p53 ubiquitination and degradation (58; data not shown), the cellular compartments may prevent PCAF degradation by this MDM2 mutant. Moreover, the nuclear export inhibitor leptomycin B did not affect MDM2-mediated PCAF degradation (Fig. 5D), supporting the idea that PCAF was degraded in the nucleus. Consistently, the resultant decrease in PCAF levels by MDM2 was mediated by the proteasomal pathway, as this decrease was reversed by the proteasome inhibitor MG132 (Fig. 5E) and ALLN (data not shown). Well correlated with these results was that the level of endogenous PCAF in p53 Ϫ/Ϫ MEF cells, which contains endogenous MDM2, was markedly lower than that in p53 Ϫ/Ϫ / mdm2 Ϫ/Ϫ MEF cells (Fig. 5F). In conclusion, these results indicate that MDM2 mediated PCAF degradation in the nucleus.
MDM2 Reduces the Half-life of PCAF-To demonstrate that MDM2 regulates the stability but not the translation of PCAF, we determined the effect of MDM2 on the proteasomal turnover FIG. 6. MDM2 shortens the half-life of PCAF in cells. A, overexpression of MDM2 results in the fast turnover of PCAF. 60% confluent H1299 cells in 100-mm plates were transfected with pCX-Flag-PCAF plasmid (3 g) alone or together with pCMV-HA-MDM2 plasmid (3 g). 40 h after transfection, cells were treated with cycloheximide (50 g/ml) and harvested at different hours after treatment. 100 g of proteins were loaded onto an SDS gel for IB using anti-Flag antibodies. This result was also repeated in p53/mdm2 double knock-out MEF cells (data not shown). B, MDM2 leads to a decrease of both free and complex forms of PCAF in cells. H1299 cells were transfected with either the PCAF plasmid alone or together with the MDM2 plasmid. Cell lysates containing 400 g of proteins prepared 48 h after transfection were used for a glycerol gradient sedimentation centrifugation. Molecular weight markers as indicated at the top of the panels were mixed with the lysates for this analysis. 30 l of fractions as indicated below the panels were used for IB analysis using antibodies against PCAF (upper panels) and MDM2 (lower panels).
of PCAF in cells. p53-deficient H1299 or p53 and mdm2 double knock-out MEF cells were transfected with the Flag-PCAF plasmid either alone or together with the HA-MDM2 plasmid. Cells were treated with cyclohexamide, a protein translation inhibitor, and harvested at different time points post-treatment for immunoblot analysis with anti-Flag antibodies. As shown in Fig. 6A, PCAF level was remarkably reduced in the presence of MDM2, and its half-life was markedly shortened by ϳ50% from greater than 4 h to ϳ2 h (Fig. 6A). MDM2 also remarkably reduced the half-life of PCAF in p53 Ϫ/Ϫ /mdm2 Ϫ/Ϫ MEF cells. 2 To further determine whether degradation of PCAF by MDM2 would affect the steady level of the free form or complex form of PCAF in cells, we performed a transient transfection of H1299 cells with PCAF alone or together with MDM2, followed by a glycerol gradient sedimentation centrifugation analysis. As shown in Fig. 6B, overexpression of MDM2 remarkably reduced both the free and complex forms of PCAF in the cells (compare the levels of PCAF between the right and left panels). These results demonstrated that MDM2 enhanced the proteasomal turnover of PCAF by ubiquitinating and degrading this protein. This degradation led to marked reduction of both free PCAF proteins and the complex form of PCAF in cells. DISCUSSION Our study demonstrates that we have identified PCAF as another ubiquitination substrate for MDM2. At least three lines of evidence support this statement. First, MDM2 ubiquitinates PCAF in vitro and in cells. Also, overexpression of MDM2 resulted in marked reduction of PCAF at the protein, but not mRNA, level. This reduction is reversed by the proteasome inhibitor MG132 and ALLN, suggesting that MDM2-led PCAF reduction is caused by the ubiquitin-proteasome system. In line with these results, endogenous PCAF levels are much lower in MDM2-containing MEF cells than in MDM2-deficient MEF cells. Interestingly, MDM2 appears to mediate PCAF degradation in the nucleus, which is consistent with a recent report showing that p53 polyubiquitination and degradation mediated by MDM2 can also occur in the nucleus (60). Finally, MDM2 significantly reduces the half-life of PCAF and the level of the free and the complex-associated PCAF molecules in cells. Thus PCAF is a new target for the E3 ubiquitin ligase MDM2.
Identification of PCAF as a substrate for MDM2 raises several interesting questions and possibilities. First, our previous study showed that MDM2 inhibited PCAF-catalyzed p53 acetylation in vitro and in cells (45). It is possible that the downregulation of PCAF-mediated p53 activation by MDM2 is partially attributed to the ability of MDM2 to ubiquitinate PCAF and to target this protein for proteasomal degradation. Also, because PCAF has been shown to form a large complex with more than 20 different proteins, including some TATA-binding protein-associated factors in cells (53,56), it is likely that MDM2 may impair the complex formation by reducing the level of PCAF (Fig. 6B). In doing so, MDM2 might suppress transcription of those genes whose expression requires the PCAFassociated complex. Alternatively, MDM2 may influence PCAF-mediated transcription by ubiquitinating this protein.
Correlated with this hypothesis is the observation that MDM2 is able to bind to and ubiquitinate PCAF that associates with a complex (Figs. 3 and 4) and to inhibit PCAF-stimulated p53 transcriptional activity by associating with the p53 responsive promoter (45), although it is still necessary to clarify whether this inhibition is caused by the association of MDM2 with p53 or PCAF ubiquitination, or both, at the promoter. Moreover, PCAF has two other homologs, PCAF-B and hGCN5 (56). It would be interesting to see if MDM2 also regulates these PCAF homologs through a ubiquitin-dependent mechanism. Finally, p300/CBP was recently shown to assist MDM2 in polyubiquitinating p53 and leading to its degradation (57). PCAF was originally identified as a p300/CBP-associated protein (50). Thus the next step would be to determine whether p300/CBP may also help MDM2 polyubiquitinate PCAF and accelerate its degradation, given our observation that PCAF appeared to be more efficiently ubiquitinated by MDM2 in cells than in vitro (Figs. 2 and 4). Of note, PCAF is a tissue-or cell-specific protein (50), and thus its regulation by MDM2 may occur merely in a tissue-or cell-specific fashion under certain pathological or physiological conditions. Addressing these remaining questions in the near future would further our understanding of the MDM2-mediated regulation of the PCAF-p53 pathway as well as other PCAF-dependent transcriptional pathways.